The perks of collaborating with a robot

A series of debt collection automation measures enable Arvato Financial Solutions’ debt collection team to work smarter and more efficiently. In Sweden, Andrea uses machine learning to assess the probability that payment will be received from a debtor, while in Norway, Tommy’s working day has changed after the addition of a robot to the team.

Andrea works with large receivables and special collection in Varberg. She handles enquiries from both private individuals and businesses.

- The working day is varied and exciting. My task is to strike a balance between what should be paid to the companies we represent and the debtor's needs on a day-to-day basis, she explains.

During the course of a day she talks personally to customers who have received a debt collection claim.

- I help people get back control of their finances, and to become free of debt in the long run, which is an important social responsibility.

Algorithm with different variables to assess probability

Andrea works with special collections in Varberg. She is clear that the customer service team and their human qualities will always be needed.

Andrea is one of the contributors to the machine learning models that has been implemented for debt collection. The models assesses the probability that payment will be received from the debtor.

- It's based on an algorithm that applies a large set of different variables such as previous payments, age, gender to assess the debtor's probability to pay.

This process was previously more manual but has now been automated. This has increased both our efficiency and stability, contributing to more consistent assessment of each individual case.


Human qualities will still be essential

Andrea believes that customer service and human qualities will always be needed.

- We need them both in combination. Based on the results from our new models, we have also identified several new areas that we can automate, and make smarter use of our resources, such as spending more time on complicated debt cases, she says.

New colleague

In Rørvik, machine learning and automation have made Tommy’s  work more effective. On behalf of the entire collection department, his role is to submit attachment requests to the bailiff's office. This previously involved a lot of manual work, but during the past year he has had the help of a new "colleague", ensuring that many of the tasks are automated. The colleague's name is Elsa, and she is a robot.


Digitalized and automated

The assessment prior to attachment has been streamlined with machine learning, and communication with the bailiff's office has been digitalised and automated. This significantly reduces the workload and assures both quality and procedures.

- Elsa is a really positive addition, and the keener we are to work with Elsa, the more efficient we become, says Tommy.

Tommy’s working day starts with manual control of cases from Elsa. Are names and addresses correct? Have payments been received during the last 24 hours? Has the debtor been in touch after the case was assigned to Elsa? Once the manual control is completed, Elsa takes over.

- Before Elsa, we printed one letter at a time, says Tommy. Everything had to be stapled, signed by a legal officer, packed in envelopes and posted. Now, it’s all done at the touch of a button. It’s pretty cool to see how much time we actually save, he explains.

With Elsa on the team, there has been more automation and less manual processes for both Tommy and the customer service team in Rørvik.

The future is even smarter

The biggest saving concerns the follow-up process. The cases used to be sent by post, but now everything is received digitally and automatically by Elsa. Before, everything had to be scanned and entered manually. Now, Elsa automatically loads it to the system and the case officers can easily follow up on their activities, preferably with reminders from Elsa of what needs to be done.

- I find it pretty fascinating that a robot can remind you of something you’ve forgotten. It’s really cool, says Tommy. 

Both Andrea and Tommy have a lot of positive things to say about smarter use of resources facilitated by machine learning and automated processes.

- We’re constantly finding new areas for improvement. In the next few years we'll have even more smart models and most likely an increase in the number manual work items that can be automated. If we work smarter, we free up time to support debtors that really need our full attention and care, which ultimately is beneficial to them, and to our customers, Andrea concludes.